Exoplanet detection methods: radial velocity, transit photometry, direct imaging

Intuition Beginner

Finding a planet around another star is like spotting a firefly circling a searchlight from kilometres away. The planet is billions of times fainter than its host star and sits right beside it, lost in the glare. Direct viewing fails for almost all of the thousands of worlds now known, so astronomers rely on indirect tricks that read the planet's influence on its star.

The radial velocity method watches the star wobble back and forth as an orbiting planet tugs it. The tug shifts the star's light slightly toward blue and red through the Doppler effect. This is how Michel Mayor and Didier Queloz found 51 Pegasi b in 1995, the first exoplanet around a Sun-like star: a gas giant whirling around its star every four days.

The transit method catches the tiny dip in starlight when a planet crosses in front of its star, like a moth dimming a lamp. NASA's Kepler telescope used this to find thousands of planets. Direct imaging goes the other way: a coronagraph masks the star's blinding light so the planet can be photographed outright, a method that works best for large, young planets orbiting far from their stars.

Visual Beginner

Method	What it watches	Best at finding	Key facility
Radial velocity	Star's colour shift (wobble)	Massive close-in planets	HARPS, ESPRESSO
Transit	Dip in star brightness	Planets of all sizes, close in	Kepler, TESS, CHEOPS
Direct imaging	The planet's own light	Large young planets far out	VLT/SPHERE, JWST

Worked example Beginner

Example 1: How deep is a transit?

When a planet crosses its star, the blocked fraction of light equals $(R_p/R_s{)}^2$, the ratio of the planet's radius to the star's radius, squared. For Jupiter ($R_p=71,492$ km) crossing a Sun-like star ($R_s=696,000$ km):

$(71492/696000{)}^2\approx 0.0106$, a 1.0 percent dip.

For Earth ($R_p=6,371$ km) crossing the same star:

$(6371/696000{)}^2\approx 0.000084$, only 84 parts per million.

The takeaway: an Earth-sized transit is about a hundred times shallower than a Jupiter-sized one, which is why stable space telescopes above the atmosphere are needed to catch them.

Example 2: How fast does the star wobble?

51 Pegasi b drags its star back and forth at about 56 metres per second, fast enough that 1995-era spectrographs caught it. An Earth twin at 1 AU around a Sun-like star produces a wobble of only about 0.09 metres per second, slower than a walking pace and right at the floor of today's best instruments.

The takeaway: massive planets on short orbits shout; small planets on wide orbits whisper, and only the most precise spectrographs can hear them.

Check your understanding Beginner

Formal definition Intermediate+

Radial velocity (Doppler spectroscopy)

A planet and its star orbit their common barycentre. The star's reflex motion has a line-of-sight component $v_r(t)$ recovered from the Doppler shift of absorption lines in the stellar spectrum. The observable is the radial-velocity curve, parameterised by the semi-amplitude $K$, the period $P$, the eccentricity $e$, and the argument of periastron $\omega$. High-resolution spectrographs ($R\gtrsim 10^5$) calibrated by iodine cells, thorium–argon lamps, or laser frequency combs deliver precisions from a few m/s in the 1990s down to roughly 0.1 m/s today (ESPRESSO). Because $v_r$ measures only the projected orbit, the method constrains $m_p\sin i$, a lower bound on the mass ^{[source pending]}.

Transit photometry

If the orbital plane is close to edge-on, the planet periodically occults the stellar disc. For a uniformly bright disc the fractional flux deficit — the transit depth — is $\delta =(R_p/R_{\star }{)}^2$, yielding the radius ratio directly. The interval between successive transits gives $P$; the duration and the shape of ingress, flat bottom, and egress constrain the impact parameter $b$, the scaled semi-major axis $a/R_{\star }$, the inclination $i$, and the limb-darkening coefficients. The geometric transit probability for a circular orbit is $P_{\mathrm{tr}}\approx R_{\star }/a$. Limb darkening curves the bottom and must be modelled for accurate radii ^{[source pending]}.

Direct imaging

Blocking the stellar point-spread function with a coronagraph or an external starshade lets the planet's own reflected or thermal light be resolved. The two fundamental limits are the diffraction-limited angular separation $\theta \approx 1.22\lambda /D$ and the raw contrast, the star-to-planet flux ratio, which ranges from $10^5$ for young wide giants to $10^{10}$ for an Earth analogue. Ground-based extreme adaptive-optics systems (VLT/SPHERE, Gemini/GPI, Subaru/SCExAO) correct atmospheric seeing in real time; space telescopes avoid it. The method favours young, self-luminous, wide-separation giant planets ^{[source pending]}.

Astrometry, microlensing, and timing

Astrometry measures the two-dimensional sky-plane reflex motion, yielding the true mass and the orbital inclination rather than $m_p\sin i$; Gaia targets micro-arcsecond precision and is expected to deliver a giant-planet census of nearby stars. Gravitational microlensing detects the transient magnification of a background star by a foreground lens system and is sensitive to cool, low-mass planets at orbital distances of a few AU, but each event is a non-repeating one-shot measurement. Pulsar timing and transit-timing variations probe multi-planet dynamics through departures from strict periodicity, with timing precision reaching nanoseconds for millisecond pulsars.

Key result: the mass–separation detectability map Intermediate+

The three principal methods are governed by a small set of analytic expressions that fix where each one is sensitive. Deriving them shows why the methods are complementary rather than redundant.

Radial-velocity semi-amplitude. For a planet of mass $m_p$ on a Keplerian orbit with period $P$, eccentricity $e$, and inclination $i$ around a star of mass $M_{\star }$, the star's line-of-sight reflex velocity oscillates with semi-amplitude

$$K=\frac{(2\pi G{)}^{1/3}}{\sqrt{1-e^2}}\frac{m_p\sin i}{(M_{\star }+m_p{)}^{2/3}}{P}^{-1/3}.$$

In the limit $m_p\ll M_{\star }$ this reduces to $K\propto m_p{P}^{-1/3}{M}_{\star }^{-2/3}\sin i$, so the signal is largest for massive planets on short orbits — exactly the hot Jupiters found first. For 51 Peg b ($m_p\sin i\approx 0.47M_J$, $P=4.23$ d, $M_{\star }\approx 1.04M_{\odot }$, $e\approx 0$) the formula gives $K\approx 55$ m/s, comfortably within 1995 reach. An Earth twin ($m_p=M_{\oplus }$, $P=365$ d, $M_{\star }=M_{\odot }$) gives $K\approx 0.09$ m/s, at the floor of present-day instruments ^{[source pending]}. Because $K$ measures only $m_p\sin i$, radial velocity returns a minimum mass; an independent transit measurement breaking the $i\approx 90^{\circ }$ degeneracy converts this into a true mass.

Transit depth and probability. For a uniform stellar disc the flux deficit is

$$\delta ={\left(\frac{R_p}{R_{\star }}\right)}^2,$$

independent of the system's distance from us. A Jupiter–Sun pair gives $\delta \approx 1.0\%$; an Earth–Sun pair gives $\delta \approx 84$ ppm. The geometric probability that a circular orbit transits is $P_{\mathrm{tr}}\approx R_{\star }/a$, so close-in planets both transit more often and repeat more frequently — a double selection bias toward short periods ^{[source pending]}.

Complementarity. Radial velocity scales with mass and $P^{-1/3}$; transit photometry scales with the squared area ratio and is further biased toward short $P$ through $P_{\mathrm{tr}}$; direct imaging scales with angular separation and planet luminosity. No single technique covers the full mass–separation plane, so a complete census stitches their overlapping sensitivity windows: radial velocity dominates at masses above a few tens of Earth masses, transits dominate at small radii and short periods, and direct imaging dominates at wide separations beyond the snow line.

Exercises Intermediate+

Advanced results Master

Radial-velocity precision floors

The achievable precision is set by the star, not the spectrograph. Stellar activity produces a stochastic jitter floor: starspots and plage modulate the line profiles as the star rotates, granulation and supergranulation drive convective blueshift fluctuations at the m/s level, and magnetic cycles impose coherent, years-long signals that mimic long-period companions. Disentangling these from genuine planetary signatures is the central technical problem of precision radial velocity. Telluric absorption bands (water vapour, molecular oxygen) imprint variable features that must be removed with atmospheric transmission models or contemporaneous airmass calibration. Laser frequency combs anchor the wavelength scale to centimetre-per-second repeatability, but the stellar noise floor — roughly 1 m/s for Sun-like stars and higher for active M dwarfs — caps the Earth-twin sensitivity. Optimal multi-Keplerian fitting uses Markov-chain Monte Carlo or nested sampling over $\{P,K,e,\omega ,T_0\}$ for each companion, marginalising over a nuisance jitter term and any long-term acceleration trend ^{[source pending]}.

Transit light-curve modelling and secondary signals

The Mandel–Agol algorithm computes the integrated flux blocked by a planet crossing a limb-darkened stellar disc, parameterised by the limb-darkening law (linear, quadratic, or Claret four-coefficient). Production pipelines — EXOFAST, batman, juliet, Allesfitter — fit the radius ratio $R_p/R_{\star }$, the scaled semi-major axis $a/R_{\star }$, the impact parameter, the period, and the limb-darkening coefficients jointly with detrending systematics for instrumental and astrophysical noise. Secondary-eclipse measurements, taken when the planet passes behind the star, isolate the dayside thermal emission; phase-curve monitoring through the full orbit maps the longitude–temperature structure and the efficiency of heat redistribution. The Rossiter–McLaughlin effect — the anomalous radial-velocity signature produced as the planet sequentially covers rotating regions of the stellar disc — yields the sky-projected spin–orbit angle $\lambda$, distinguishing quiescent disc migration (well-aligned orbits) from high-eccentricity scattering or Kozai–Lidov migration (misaligned or retrograde orbits) ^{[source pending]}.

Atmospheric characterisation and retrieval

Transmission spectroscopy during transit probes the terminator atmosphere; emission spectroscopy during secondary eclipse probes the dayside. The forward model — a radiative-transfer calculation over pressure-temperature profiles, chemical abundances, and cloud opacities — is inverted against the observed spectrum by retrieval codes such as PLATON, ATMO, NEMESIS, and petitRADTRANS, which use Bayesian or nested-sampling exploration of the parameter space. JWST/NIRSpec and HST/WFC3 have retrieved water vapour, carbon dioxide, methane, and sulphur dioxide in hot-Jupiter and sub-Neptune atmospheres; the detection of sulphur dioxide in WASP-39b is direct evidence of photochemistry driven by stellar ultraviolet flux. For temperate rocky planets around M dwarfs, including the TRAPPIST-1 system, the open question is whether any atmosphere survives the intense X-ray and ultraviolet emission of the host star during its prolonged pre-main-sequence phase ^{[source pending]}.

Direct-imaging contrast and future facilities

A contrast curve gives the $5\sigma$ minimum detectable planet-to-star flux ratio as a function of angular separation. Current extreme adaptive-optics coronagraphs (VLT/SPHERE, Gemini/GPI, Subaru/SCExAO) reach raw contrasts of roughly $10^{-6}$ at separations of 0.1–0.5 arcsec in the near-infrared, sufficient for young self-luminous giants such as the four-planet HR 8799 system and $\beta$ Pictoris b. JWST coronagraphy extends this into the thermal mid-infrared where cooler planets peak. The Nancy Grace Roman Space Telescope coronagraph technology demonstrator and the proposed Habitable Worlds Observatory and LUVOIR concepts target reflected-light terrestrial planets at contrasts of $\sim 10^{-10}$, a three-to-four-order-of-magnitude gain requiring active wavefront sensing, deformable mirrors, and external starshades flying in formation with the telescope. ARIEL will undertake a statistical transit-spectroscopy survey of roughly a thousand warm planets, building the demographic baseline for atmospheric classification ^{[source pending]}.

System architecture and survey strategy

Multi-planet systems populate a continuum from compact resonant chains (TRAPPIST-1, Kepler-90) to dynamically hot, sparse configurations hosting a single dominant giant. Mutual Hill-separation criteria and long-term N-body integrations bound the architectures that remain stable over Gyr timescales, separating packed systems teetering on the edge of instability from sparse ones with ample dynamical room. Transit-survey yield optimisation trades sky coverage against cadence and photometric precision; habitable-zone surveys around M dwarfs exploit both the favourable $R_{\star }/a$ ratio and the short habitable-zone periods to maximise transit probability and repeat frequency. Radial-velocity follow-up of these same targets is limited by the stellar jitter floor imposed by granulation, supergranulation, and magnetic activity, which collectively make the detection of Earth-mass planets in M-dwarf habitable zones achievable but demanding.

Connections Master

This unit deepens the detection half of the exoplanet programme surveyed in the overview unit 28.05.01, which introduced the methods qualitatively and turned to habitability; the formal derivation of the semi-amplitude $K$ and the transit depth $\delta$ here is the analytic core beneath that qualitative account. The habitability thread continues downstream, where the demographic data these methods produce — occurrence rates as functions of planet radius, mass, and orbital period — feed directly into the habitability and biosignature analysis of the successor unit, a link registered as a proposed hook in the unit metadata.

Every detection method inherits its stellar calibration from stellar astrophysics. Transit radii scale linearly with $R_{\star }$ and radial-velocity masses scale with $M_{\star }^{2/3}$, so the precision of a detected planet's properties is bounded by the precision of the host-star parameters. Asteroseismology, treated in the stellar-structure and stellar-evolution units 28.02.01, 28.02.02 pending, supplies the radius, mass, and age constraints that propagate into exoplanet characterisation; the synergy was demonstrated at scale by Kepler, which returned asteroseismic solutions for thousands of planet-host giants. Comparative planetology with the solar system 28.01.01, 28.01.02 pending grounds the interpretation of exoplanet interiors and atmospheres, providing the in-situ ground truth that calibrates the remote-sensing models.

The inferential machinery extends beyond astronomy. Keplerian orbit fitting, transit-timing-variation analysis, and atmospheric retrieval all rely on Bayesian and nested-sampling techniques developed in the statistics and data-science domain, while occurrence-rate estimation from a censored, detection-efficiency-limited sample is a problem in survival analysis. The radiative-transfer kernels inside the atmospheric retrieval codes inherit line-by-line opacities from Earth-atmosphere and planetary-atmosphere work, so the unit is a consumer of both statistical inference and atmospheric radiative transfer.

Historical and philosophical context Master

The 1995 discovery of 51 Pegasi b by Mayor and Queloz — a Jupiter-mass planet on a 4.2-day orbit — overturned the expectation, shaped entirely by the solar system, that gas giants form only in the cold outer regions of a disk. The radial-velocity signal of roughly 55 m/s was large enough for the ELODIE spectrograph at the Haute-Provence Observatory, and the result was rapidly confirmed by Marcy and Butler. The discovery was recognised with the 2019 Nobel Prize in Physics.

The transit method opened atmospheric studies once the first transiting planet, HD 209458 b, was detected independently by Charbonneau et al. and Henry et al. in 1999–2000. The NASA Kepler mission, launched in 2009 and described by Borucki et al. in Science (2010), monitored roughly 150 000 stars with photometric precision better than 50 ppm and established that small planets are the most common outcome of planet formation. The Transiting Exoplanet Survey Satellite extended transit detection to an all-sky sample of bright, nearby stars, and CHEOPS was launched to refine the radii of known planets for bulk-density determination.

Direct imaging matured in 2008 with the nearly simultaneous announcement of three companions orbiting HR 8799 by Marois et al. and the detection of Fomalhaut b by Kalas et al., the first resolved images of planetary-mass objects orbiting normal stars. These systems remain the benchmarks for high-contrast imaging and for the migration and formation theory required to explain self-luminous giants at tens of astronomical units.

Earlier milestones frame the field. Wolszczan and Frail (1992) detected planets around the millisecond pulsar PSR B1257+12 by pulse-timing, the first confirmed exoplanets of any kind, though around a stellar remnant rather than a main-sequence star. The Campbel–Walker–Yang (1988) radial-velocity detection around $\gamma$ Cephei, confirmed by Hatzes et al. in 2003, may predate 51 Peg b but was not secured at the time. The interplay of pulsar timing, radial velocity, transit photometry, and direct imaging across three decades converted the existence of other worlds from speculation into a statistical discipline.

Bibliography Master

1. Mayor, M. and Queloz, D. (1995). "A Jupiter-mass companion to a solar-type star." Nature, 378, 355–359.

2. Wolszczan, A. and Frail, D. A. (1992). "A planetary system around the millisecond pulsar PSR1257 + 12." Nature, 355, 145–147.

3. Charbonneau, D., Brown, T. M., Latham, D. W., and Mayor, M. (2000). "Detection of planetary transits across a Sun-like star." Astrophysical Journal Letters, 529, L45–L48.

4. Henry, G. W., Marcy, G. W., Butler, R. P., and Vogt, S. S. (2000). "A transiting '51 Peg-like' planet." Astrophysical Journal Letters, 529, L41–L44.

5. Borucki, W. J. et al. (2010). "Kepler Planet-Detection Mission: Introduction and First Results." Science, 327, 977–980.

6. Marois, C., Macintosh, B., Barman, T., et al. (2008). "Direct imaging of a multi-planet system around the star HR 8799." Science, 322, 1348–1352.

7. Kalas, P., Graham, J. R., Chiang, E., et al. (2008). "Optical images of an exosolar planet 25 light-years from Earth." Science, 322, 1345–1348.

8. Perryman, M. (2018). The Exoplanet Handbook (2nd ed.). Cambridge University Press.

9. Carroll, B. W. and Ostlie, D. A. (2017). An Introduction to Modern Astrophysics (2nd ed.). Pearson, Chapter 23.

10. Winn, J. N. and Fabrycky, D. C. (2015). "The Occurrence and Architecture of Exoplanetary Systems." Annual Review of Astronomy and Astrophysics, 53, 409–447.

11. Fischer, D. A., Anglada-Escude, G., Arriagada, P., et al. (2014). "Star-Planet Connection I: Overview and Exoplanet Detection Techniques." Protostars and Planets VI, 715–738.

12. Mandel, K. and Agol, E. (2002). "Analytic Light Curves for Planetary Transit Searches." Astrophysical Journal Letters, 580, L171–L175.